Re: Why SyncRepWakeQueue is not static?

On Wed, Mar 25, 2015 at 10:07 PM, Tatsuo Ishii <ishii@postgresql.org> wrote:
>>> Fix committed/pushed from master to 9.2. 9.1 declares it as a static
>>> function.
>>
>> Er, is that a good idea to back-patch that? Normally routine specs are
>> maintained stable on back-branches, and this is just a cosmetic
>> change.
>
> I'm not sure if it's a cosmetic change or not. I thought declaring
> to-be-static function as extern is against our coding
> standard. Moreover, if someone wants to change near the place in the
> source code in the future, changes made to head may not be easily back
> patched or cherry-picked to older branches if I do not back patch it.

True.  But if any third-party code calls that function, you just broke
it.  I don't think keeping the back-branches consistent with master is
a sufficiently good reason for such a change.

-- 
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company